Cyclic AMP Receptor Protein Regulates cspE, an Early Cold-Inducible Gene, in Escherichia coli

Abstract

cspE, a member of the cspA family of cold shock proteins in Escherichia coli, is an early cold-inducible protein. The nucleic acid melting ability and transcription antiterminator activity of CspE have been reported to be critical for growth at low temperature. Here, we show that the cyclic AMP receptor protein (CRP), a global regulator involved in sugar metabolism, upregulates cspE in E. coli. Sequence analysis of the cspE upstream region revealed a putative CRP target site centered at −61.5 relative to the transcription start. The binding of CRP to this target site was demonstrated using electrophoretic mobility shift assays. The presence of this site was shown to be essential for P_cspE activation by CRP. Mutational analysis of the binding site indicated that the presence of an intact second core motif is more important than the first core motif for CRP-P_cspE interaction. Based on the promoter architecture, we classified P_cspE as a class I CRP-dependent promoter. This was further substantiated by our data demonstrating the involvement of the AR1 domain of CRP in P_cspE transcription. Furthermore, the substitutions in the key residues of the RNA polymerase α-subunit C-terminal domain (α-CTD), which are important for class I CRP-dependent transcription, showed the involvement of 265 and 287 determinants in P_cspE transcription. In addition, the deletion of crp led to a growth defect at low temperature, suggesting that CRP plays an important role in cold adaptation.

INTRODUCTION

Escherichia coli K-12 contains nine paralogs of CspA, CspA to CspI, collectively known as the CspA family of cold shock proteins (CSPs). Members of this family are small proteins consisting of a nucleic acid binding domain called the cold shock domain (CSD), one of the most evolutionarily conserved domains found in various life forms, including bacteria, plants, and animals (19, 47). In spite of the high degree of similarity among them, only five (cspA, cspB, cspE, cspG, and cspI) are induced during cold shock (20, 42). cspC is constitutively expressed at 37°C, while cspD is induced during starvation (49).

CspA and some of its homologues share nucleic acid melting ability (25) and transcription antitermination activity (3), two functions that are important for cold adaptation (35). A balanced level of these proteins seems to be important for cold adaptation (48). Among the nine members of this family, CspE performs functions which are important during both cold shock and diverse physiological conditions, including chromosome partitioning/condensation (24, 50), growth in cold (30), UV sensitivity (29), and the regulation of various cold-induced genes (36).

cspE expression has been found to be regulated in a growth phase-dependent manner (2). Earlier reports suggested that interplay exists between the cellular metabolic status and cold shock response (13, 26, 43). Additionally, global gene expression profiling has indicated that many cold-induced genes are subject to catabolite repression (18, 34). Cyclic AMP (cAMP) receptor protein (CRP), also known as catabolite activator protein (CAP), is responsible for the regulation of genes involved in various metabolic pathways in Escherichia coli (28). The CRP target site is a palindromic sequence in which two conserved core motifs, having the 5-bp sequence TGTGA, are separated by a 6-bp spacer. CRP-regulated promoters are classified into three classes depending on the target site location (7). For class I promoters, the CRP target site is located upstream of the RNA polymerase (RNAP) site, where the sites for CRP and RNAP are on the same face of the DNA helix (centered at −61.5 or farther upstream). For class II promoters, the CRP target site overlaps the RNAP site (−41.5), while class III promoters require additional factors for activation. Different regions on CRP were identified based on their interaction with RNAP depending on the specific class of CRP-dependent promoter (21). Activating region 1 (AR1) of CRP, composed of amino acid residues 156 to 164, interacts with a specific surface determinant, 287, from the C-terminal domain of the alpha subunit (α-CTD) of RNAP in both class I and class II promoters (21). The 265 determinant of α-CTD interacts with AT-rich sequences in DNA, particularly with the UP element, in CRP-dependent promoters. The interaction at class II CRP-dependent promoters also involves interaction between a second region, called AR2, of CRP (H19, H21, E96, and K101) and the N-terminal domain (α-NTD) of the RNAP α subunit (21). A third surface patch, AR3, of CRP (amino acid residues 52 to 55 and 58) recently has been shown to be important in CRP-dependent class II promoter activity (7, 21).

The present study provides further evidence for possible cross-talk between cold shock and cellular metabolic regulation. We show that CRP regulates cspE, an early cold-inducible gene in E. coli, by a class I CRP-dependent mechanism. Further, the presence of CRP was found to be critical for growth at low temperature. Our results suggest for the first time that CRP, a global regulator, plays an important role in cold adaptation.

MATERIALS AND METHODS

Bacterial strains, plasmids, and primers.

The strains and plasmids used in this study are given in Table 1. The primers used in this study are given in Table 2. The growth medium used was either Luria-Bertani (LB) broth or LB agar plates. Growth was monitored by determining the optical density at 600 nm (OD₆₀₀). Antibiotics were included in the medium per the requirements.

Table 1.

List of strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
Strains
M182	E. coli K12 Δ(lacIPOZY)×74 galK galU strAcrp+	9
M182Δcrp	Δcrp derivative of M182	8
BL21(λDE3)	E. coli B F−dcm ompThsdS(rB− mB−) gal λ(DE3); encodes T7 RNAP under the control of lacUV5 promoter	Novagen Inc.
AG1	endA1 recA1 gyrA96 thi-1 relA1 glnV44 hsdR17(rK− mK+)	Stratagene Inc.
JW5702	AG1, pCA24N-crp	27
Plasmids
pEU720	R100 origin; Spcr/Strr; 15.2 kb; single-copy plasmid	14
pAM1956	Kanr, 10.0 kb, low-copy-no. plasmid	51
pEUcspEgfp312	4.2-kb EcoRI fragment in pEU720 replaced with an ∼1-kb EcoRI cspE-gfp312 fragment comprising the −306 to +6 cspE region fused to translatable gfp (see Materials and Methods)	This study
pEUcspEgfp380	4.2-kb EcoRI fragment in pEU720 replaced with an ∼1-kb EcoRI cspE-gfp380 fragment comprising the −306 to +72 cspE region fused to the gfp ORF (see Materials and Methods)	This study
pEUcspEgfp48	Same as pEUcspEgfp312, except first core motif (−69 to −65 region) was deleted from CRP target site in the cspE upstream region (see Materials and Methods)	This study
pEUcspEgfp60	Same as pEUcspEgfp312 but with cspE having three point mutations in the first core motif [GC (−68), GC(−66), and AC(−65) of the CRP target site (see Materials and Methods)	This study
pCA24N	colE1 origin; Cmr; 5.24 kb; lacIq	27
pCA24N-crp	Cmr; lacIq; pCA24N PT5-lac::crp+	27
pDCRP (and derivatives)	colEI origin; Amr; wild-type crp gene (and derivatives) cloned into pBR322 with crp coding sequence on HindIII-EcoRI fragment	44
pDU9	colEI origin; Amr; pDCRP with HindIII-EcoRI fragment carrying crp replaced by M13mp8 polylinker	44
pLG339	pSC101 origin; Kanr, Tcs; 6.2 kb; low-copy-no. (6-8) cloning vector	41
pDW300	pLG339 derivative carrying crp gene	44
pDW301	pLG339 derivative carrying H159L mutant allele of crp gene	44
pHA5	Derived from pBR322; insert of 3.6-kb BamHI fragment containing crp gene and promoter	1
pXZCRP	f1 origin; Amr; multicopy plasmid derived from pHA5 having crp gene	53
pXZ180K	pXZCRP with R180K mutant allele of CRP	17
pXZ180L	pXZCRP with R180L mutant allele of CRP	17
pHTf1a (and derivatives)	f1 origin; Amr; plasmid carrying rpoA encoding RNAP α subunit (and derivatives carrying alanine substitutions at positions 165, 261, 265, and 268)	15
pREIIa (and derivatives)	colE1 origin; Amr; plasmid carrying rpoA encoding RNAP α subunit (and derivatives carrying alanine substitutions at positions 287, 294, and 296)	15

Table 2.

Primers used in this study

Primer	Sequence (5′-3′)a
SU23R	ATGCCTGCAGGTCTGAATTCTTATT
SU25F	GGTACCTACAACAATGGCAATGTGT
SU26F	GGTAACGTTAAGTGGATGAGTAAAGGAGAAGAA
SU27R	TTCTCCTTTACTCATCCACTTAACGTTACCTTTA
SU29F	ACTGGTAACCGACACACGAGCTCGGTACCCG
SU30R	CGGGTACCGAGCTCGTGTGTCGGTTACCAGT
VRS2F	GTCTGGAATTCACCGCTGGCG
SU40F	TAAGTAACATCAAAAATAACGCGACTTTTATCACTTTTTAGTAAAGTTAC
SU41R	GTAACTTTACTAAAAAGTGATAAAAGTCGCGTTATTTTTGATGTTACTTA
SU43R	AAATGCTGTGTCGGTTACCAGTACACCAATTGTGGTACGC
SU44F	GTAAAGTTACACTGGACAAAGCGTACCACAATTGGTGTACTGGTAACCGA
SU45R	TCGGTTACCAGTACACCAATTGTGGTACGCTTTGTCCAGTGTAACTTTAC
SU48F	TAAAATAAGTAACATCAAAAATAACTTTTATCACTTTTTAGTAAAGTTAC
SU49R	GTAACTTTACTAAAAAGTGATAAAAGTTATTTTTGATGTTACTTATTTTA
SU60F	TAAGTAACATCAAAAATAACGCGACTTTTATTTTAGTAAAGTTACACTGG
SU61R	CCAGTGTAACTTTACTAAAATAAAAGTCGCGTTATTTTTGATGTTACTTA

^aThe EcoRI restriction enzyme sites are underlined.

The fusion fragments cspE-gfp312, cspE-gfp380, cspE-gfp48, and cspE-gfp60 were synthesized using an overlap extension PCR method as described by Uppal et al. (42). The cspE fragment was amplified using VRS2F (forward flanking) and SU30R (overlapping reverse) primers with pEUcspE (42) as the template. Translatable gfp-mut2 was amplified using SU29F (overlapping forward) and SU23R (reverse flanking) primers with pAM1956 (51) as the template. The two overlapping fragments, having complementarity at the 3′ end, were fused to obtain the cspE-gfp312 fragment. cspE-gfp380 was synthesized in a similar manner using SU26F and SU27R as the overlapping primers. Similarly, cspE-gfp48 and cspE-gfp60 were synthesized using SU48F/SU49R and SU60F/SU61R as the overlapping primers, respectively, and VRS2F and SU23R as flanking primers with pEUcspEgfp312 (see below) as the template. Both VRS2F and SU23R, the flanking primers, contained EcoRI sites (Table 2). Plasmids pEUcspEgfp312, pEUcspEgfp380, pEUcspEgfp48, and pEUcspEgfp60 were constructed by replacing the 4.2-kb EcoRI fragment from pEU720 (14) with ∼1.0-kb EcoRI cspE-gfp312, cspE-gfp380, cspE-gfp48, and cspE-gfp60 fragments, respectively. For transcriptional fusion (pEUcspEgfp312), the −306 to +6 region, including P_cspE and 6 bases of the cspE untranslated region (UTR), was fused to translatable gfp. For translational fusion (pEUcspEgfp380), the −306 to +72 region, containing P_cspE, 42 bases of the UTR (including the translational signals of cspE) and sequence coding for the initial 10 codons of the cspE open reading frame (ORF), was fused to the gfp ORF.

Green fluorescent protein (GFP) fluorescence intensity assay.

Freshly diluted cultures (1:100) were grown at 37°C in a conical flask with constant shaking. Both GFP-mut2 fluorescence (excitation, 480 nm; emission, 510 nm) and absorbance at 600 nm were measured (OD₆₀₀) in a multiwell plate reader (Infinite200; Tecan, United Kingdom). For the plate assay, freshly diluted (1:100) cultures were grown in 96-well plates without shaking. Unless otherwise mentioned, the flask method was used. The intensity from plain LB broth was used for blank correction. Path-length correction was carried out per the manufacturer's instructions. Specific fluorescence intensities (SFI) were calculated by normalizing fluorescence values against the OD₆₀₀ for the respective wells. Since the GFP-mut2 variant is highly stable and nontoxic to the cells, the accumulation of the fluorescence serves as a reporter of the transcription/translation initiation in the cells (52).

His-tagged CRP purification.

E. coli BL21(λDE3) pLysS containing pCA24N-crp plasmid (27) (Table 1) was used to obtain the purified CRP using a protocol essentially as described in Hire et al. (23). CRP protein was induced by the addition of 0.2 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and the bound protein was eluted using 50 to 500 mM imidazole gradient. The pure protein fractions were pooled, concentrated, and exchanged in TE buffer (20 mM Tris-HCl pH 8.0, 1 mM EDTA) using an Amplicon 5K column (Amicon).

Electrophoretic mobility shift assay (EMSA).

The DNA samples (PCR products) were mixed with purified His-tagged CRP (at appropriate concentrations) in 10 μl 1× binding buffer (10 mM KCl, 4 mM Tris HCl [pH 8.0], 1 mM EDTA, 0.1 mM dithiothreitol [DTT], 200 μM cyclic AMP, 10 μg/ml bovine serum albumin) and incubated at 37°C for 10 min. Samples were mixed with loading dye (0.25% bromophenol blue, 0.25% xylene cyanol FF, 15% Ficoll 400) and loaded on a 10% nondenaturing polyacrylamide gel containing 200 μM cyclic AMP (Sigma). The gel was electrophoresed at constant voltage (15 V/cm), stained in 1× SYBR green I nucleic acid gel stain (Roche), and visualized under UV light using a gel documentation system (Syngene, United Kingdom). The deletion fragments used for EMSA were amplified using pEUcspE (42) as a template, and the primers used were the following: VRS2F and SU43R for the fragment spanning −300 to +12, SU25F and SU43R for −180 to +12, SU40F and SU43R for −88 to +12, SU40F and SU45R for −88 to +2, and SU44F and SU43R for −48 to +12. The smallest fragment spanning the −88 to −38 region was obtained by mixing the overlapping primers SU40F and SU41R. The mutant fragments were synthesized using the corresponding mutant forward primers (the same as SU40F except for the deletions/point mutations) and SU43R, the reverse primer, with pEUcspE as the template.

Growth curve using automated optical density-monitoring system, Bioscreen C.

Freshly diluted (1:100) cultures (100 μl) were grown in a honeycomb 100-well plate (Bioscreen; OY Growth Curves, Finland) in a Bioscreen C machine with continuous shaking at 37°C. Absorbance readings at 600 nm (wideband range) were taken every 30 min. Plain LB broth was used for blank corrections. OD₆₀₀ values obtained in this machine are not directly comparable to those from a standard spectrophotometer due to a shorter path length (∼0.38 cm) of the vertical optical beam through the culture (100 μl) contained in the wells of honeycomb plates. The path-length correction factor has been calculated from the slope of a standard curve plotted between values from a spectrophotometer and honeycomb plate using a Bioscreen machine for a culture across a range of ODs. Total numbers of CFU were measured by plating appropriate dilutions of the saturated cultures.

RESULTS

CRP positively regulates P_cspE.

The sequence analysis of the cspE upstream region revealed a putative CRP target site, spanning the region of −72 to −51, centered at −61.5 relative to the transcription start site (TSS) (Fig. 1A and B). To investigate the role of CRP in regulating cspE, both M182 and M182Δcrp strains were transformed with pEUcspEgfp312, a transcriptional fusion. The specific fluorescence intensities (SFI), defined by fluorescence signal per cell, and culture OD₆₀₀, measured during growth at 37°C, were plotted against time (Fig. 1C). Results showed that the SFI levels were ∼2-fold lower in crp knockout cells than in the wild-type (WT) cells. Further, the complementation of M182Δcrp cells with a multicopy crp plasmid (pDCRP) resulted in a 10-fold increase in the SFI level (Fig. 1D). These results indicated that CRP regulates P_cspE, and this regulation is proportional to the crp copy number.

Fig. 1.

(A) Nucleotide sequence of cspE upstream region showing the CRP target site, P_cspE, and the transcription start (+1). (B) The 22-base consensus CRP target site sequence and the CRP target site in P_lac and P_cspE are aligned against each other. The numbering of the nucleotide positions in the target site is essentially the same as that used by Ebright et al. (11). The dyad axis of symmetry is shown by a forward slash (between nucleotide positions 11 and 12). The first conserved core motif is from nucleotide positions 4 to 8, and the second core motif is from positions 15 to 19. Nucleotide positions 9 to 14 are termed as spacer region in the CRP target site. The position of primary kinks is shown by vertical lines (between 6 and 7 in first motif and 16 and 17 in second motif). The positions of the dyad axes of the target sites in P_cspE (−61.5) and P_lac (−60.5) are mentioned in parentheses. The residues identical to the consensus are shown in boldface and are italicized, and the two core motifs are underlined. (C to E) Plots of specific fluorescence intensity (SFI), expressed in arbitrary units (AU), against time for M182 and M182Δcrp cells harboring pEUcspEgfp312, the transcriptional fusion (C), M182Δcrp/pEUcspEgfp312 cells harboring pDU9 (control) and pDCRP (WT CRP) multicopy plasmids (D), and M182, M182Δcrp, M182Δcrp/pDU9, and M182Δcrp/pDCRP cells harboring pEUcspEgfp380, the translational fusion (E). OD₆₀₀ values are plotted on the secondary axes in panels C and D. The experiment was repeated three times, and the SFI data correspond to the means from three independent repeats with error bars showing the standard errors. For the OD curve, only a representative graph is presented.

Furthermore, results from translational fusion (pEUcspEgfp380) assays (Fig. 1E) showed a similar decrease in the SFI level (∼2-fold) in crp knockout cells compared to the level in the wild-type cells. In addition, the complementation of M182Δcrp cells with multicopy crp plasmid (pDCRP) restored the translational fusion activity to more than normal. These observations demonstrate that CRP-mediated P_cspE regulation is accompanied by a concomitant increase in the level of translation in the cells.

CRP binding site is centered at −61.5 in the cspE upstream region.

The CRP target site is a palindromic sequence, where two conserved core motifs, having the 5-bp sequence TGTGA, are separated by a 6-bp spacer (Fig. 1B). EMSA using a 312-base fragment (−300 to +12) from the cspE upstream region and purified CRP revealed the binding of CRP to this region as evident by an additional, slow-moving band in the presence of CRP (Fig. 2A, lane 2). Further, DNA fragments having deletions upstream of −88 (Fig. 2A, lanes 4, 6, 8, and 10) and downstream of −38 (Fig. 2A, lane 8) showed binding to CRP, while a deletion downstream of −88 (lane 12) abolished CRP binding. These results mapped the CRP binding site to the region of −88 to −38 (Fig. 2C), which overlaps the predicted CRP target site (−72 to −51). Further analysis using the fragment spanning the region of −88 to +12 revealed that CRP binding to this sequence is concentration dependent (Fig. 2B). Deletions of either of the two core motifs as defined in Fig. 1B, or both of them together, abolished CRP binding (Fig. 3A, lane 3 to 8), thereby validating that CRP binds to the putative target site sequence, which is centered at −61.5 (relative to the TSS) in the cspE upstream region.

Fig. 2.

Electrophoretic mobility shift assays (EMSA) showing binding of CRP to the cspE upstream region. (A) DNA fragments containing deletions in the cspE upstream region (between −300 to +12) were electrophoresed in a polyacrylamide gel (10%) with 200 μM cAMP, with and without CRP (90 nM) in the binding buffer. The coordinates of the DNA fragments are shown on the top of the gel. The positions of the bound fragments are marked with arrows. (B) EMSA showing the binding of various concentrations of CRP (0 to 240 nM) to the 100-bp (−88 to +12) cspE fragment. (C) A line diagram depicting the results of CRP binding to the cspE upstream region (−300 to +12) carried out as described for panel A. The presence of CRP binding is depicted by a plus sign, and no binding is shown by a minus sign.

Fig. 3.

(A to C) EMSA showing the effect of deletions and mutations in the CRP target site contained in the 100-bp cspE fragment. The changes introduced in the target site are mentioned at the top of each lane. (D) A line diagram depicting the results of EMSA shown in panels A to C. The binding efficiency was measured qualitatively by analyzing the intensity of the bound fragment as a percentage of the bound wild-type fragment for equal amounts of starting DNA (150 ng) with a known single concentration of CRP (90 nM). The conserved core motifs in the CRP site are underlined. The notations for binding efficiency are, in decreasing order of efficiency, a large plus sign, a small plus sign, and a minus sign. (E) A plot of SFI against time for M182 and M182Δcrp cells harboring pEUcspEgfp312, pEUcspEgfp48, or pEUcspEgfp60. The plate fluorescence assay was used to measure cspE expression. The data correspond to the means from three independent repeats, and the error bars show the standard errors.

Sequence integrity of the second core motif is more important for CRP binding.

The residues important for the binding of CRP to the target site in P_cspE were identified by introducing individual point mutations in the target site followed by gel shift analysis (Fig. 3). Here, the terms “first core motif” and “second core motif” are used as defined in the legend to Fig. 1B. Mutations in the left region flanking the first core motif (Fig. 3A, lane 10), single point mutations at the 5th G (Fig. 3A, lane 12), 7th G (Fig. 3B, lane 4), and 8th A (Fig. 3B, lane 6) of the first core motif and the 15th T of the second core motif (Fig. 3B, lane 10) are tolerated moderately, as evidenced by the decreased intensity of the shifted band. The simultaneous presence of all three point mutations in the first core motif (Fig. 3B, lane 8) and the single point mutations at the 16th C (Fig. 3B, lane 12) and 18th C (Fig. 3C, lane 4) of the second core motif are tolerated poorly. The simultaneous presence of all three point mutations in the second core motifs (Fig. 3C, lane 6) and all six point mutations in both motifs combined (Fig. 3C, lane 8) did not show any shifted band. Taken together, these data demonstrated that the changes are tolerated relatively more poorly in the second core motif than in the first core motif, suggesting that the presence of an intact second core motif is more important for CRP binding to P_cspE.

Partial deletion of CRP target site leads to loss of P_cspE activation by CRP.

Whether the presence of the target site is a prerequisite for CRP-mediated P_cspE activation was analyzed by introducing the following cspE-gfp transcriptional fusions: pEUcspEgfp48, having a deletion of the first core motif (deleted of positions −69 to −65); pEUcspEgfp60, having three point mutations in the first core motif [GC(−68), GC(−66), and AC(−65)]; and pEUcspEgfp312, the wild-type control in both M182 and M182Δcrp strains. In line with earlier results (Fig. 1C), wild-type P_cspE activity was lower (∼2.5-fold) in M182Δcrp than in M182 (Fig. 3E). However, for both pEUcspEgfp48 and pEUcspEgfp60, the SFI level was negligible and the differential regulation across the two strains disappeared, demonstrating that CRP-mediated P_cspE activation is completely abolished in the absence of CRP binding.

Effect of defective DNA binding domain of CRP on P_cspE activation.

To test the effect of a defective DNA binding domain of CRP on P_cspE activation, multicopy plasmids pXZCRP180K (R180K allele), pXZCRP180L (R180L allele), and pXZCRP (wild-type crp) were introduced separately into M182Δcrp cells harboring pEUcspEgfp312. cspE expression was almost 2-fold less in the presence of pXZCRP180K and 9-fold less in the presence of pXZCRP180L than that of the wild type (Fig. 4A). The larger reduction seen for the R180L mutant could be explained by the severe defect in the DNA binding ability of this allele (17). These results confirmed that the DNA binding ability of CRP is essential for P_cspE activation.

Fig. 4.

Role of different functional domains of CRP in P_cspE activation. A plot of SFI against time for M182Δcrp/pEUcspEgfp312 cells harboring different mutants alleles of CRP on a multicopy plasmid. The data correspond to the means from three independent repeats, and the error bars show the standard errors. (A) Complementation with plasmids carrying WT CRP (pXZCRP) and CRP mutants with defective DNA binding domains (pXZCRP180K and pXZCRP180L). (B) Complementation with plasmids carrying no CRP (pDU9; only plasmid), WT CRP (pDCRP), and CRP mutants with a defective AR1 domain (pDCRPHL159), a defective AR2 domain (pDCRPKE101), and both (pDCRPHL159KE101). (C) Complementation with plasmids carrying no CRP (pDU9) or WT CRP (pDCRP) and CRP mutants with an improved AR3 domain (pDCRPKN52) and an improved AR3 domain with defective AR1/AR2 domains (pDCRPHL159KE101KN52).

Effect of CRP with modified AR1/AR2 and AR3 on P_cspE activation.

To test the involvement of different domains of CRP in P_cspE activation, plasmids pDCRPHL159 (with the H159L allele having a defective AR1 domain), pDCRPKE101 (with the K101E allele having a defective AR2 domain), pDCRPHL159KE101 (with defective AR1 and AR2 domains), pDCRPKN52 (with the K52N allele and an improved AR3), pDCRPHL159KE101KN52 (K52N H159L K101E CRP allele), pDCRP (wild-type control), and pDU9 (plasmid control without crp) were transformed separately into M182Δcrp cells harboring pEUcspEgfp312. Results showed that K101E mutation in CRP did not affect P_cspE activation, while both H159L and H159L K101E alleles adversely affected P_cspE activation (Fig. 4B). In the presence of the K52N allele, P_cspE activation decreased by ∼70%, while the H159L K101E K52N allele did not show P_cspE activation (Fig. 4C). These results indicate that the AR1 domain of CRP, not the AR2 domain, is critical for P_cspE activation, while an improved AR3 domain did not contribute to the activation.

Further, in the presence of a defective AR1 domain, the P_cspE activity was low (Fig. 4B) while cell growth was similar to that of the wild type (data not shown). This corroborated our earlier conclusion (Fig. 1C) that the difference in the P_cspE activity in the crp deletion mutant and the wild-type strain is due to altered CRP availability, not due to the different growth rates of these two strains.

Effect of mutations in α-CTD of RNAP on P_cspE transcription.

To investigate the role of α-CTD in P_cspE activation, M182Δcrp/pEUcspEgfp312 cells were cotransformed with plasmids harboring either wild-type crp (pDW300) or H159L crp mutant (pDW301) and a plasmid having a defective rpoA gene coding for a mutant α-CTD (Table 1). A series of rpoA mutants having alanine substitutions at specific residues representing the 265 determinant (R265, N268, N294, and G296), the 261 determinant (V261), and the 287 determinant (V287) of α-CTD were used. One mutant of α-NTD (E165) also was included in this study. In the presence of wild-type CRP, the V287 mutation decreased P_cspE transcription by the maximum level (∼64%) (Fig. 5). Alanine substitutions in residues R265, N268, N294, and G296 caused a moderate decrease in promoter activity (in the range of 20 to 36%), while the substitution at the V261 residue increased P_cspE transcription by 20%. In the presence of the H159L crp mutant, the alanine substitutions at V287, V261, and N294 did not cause any significant change in transcription, whereas R265, N268, and G296 showed a decrease ranging between 24 and 50%. Alanine substitutions at residue E165 in α-NTD did not make any difference to P_cspE transcription. These results indicated the involvement of different residues of α-CTD to various degrees, with V287 affecting P_cspE transcription most severely.

Fig. 5.

Bar diagram showing SFI against the positions of alanine substitutions for a series of α-CTD mutants of RNAP for M182Δcrp/ pEUcspEgfp312 cells containing either pDW300 (WT CRP) or pDW301 (H159L mutant CRP) with rpoA (α-CTD) mutants on a third plasmid. The rpoA alleles carrying the alanine substitutions at positions E165, V261, R265, and N268 are present on pHTf1a (wt1), and V287, N294, and G296 are present on pREIIa (wt2). The absolute SFI values (in AU) for wt1 and wt2 are ∼182,666 and 178,531, respectively, in the presence of WT CRP and 54,280 and 55,221, respectively, in the presence of the H159L allele. For each α-CTD mutant, the cspE transcription level was expressed as a percentage of SFI from the wild type α-CTD allele, and the value at 20 h after inoculation was plotted. The plate fluorescence assay was used to measure cspE expression. The data correspond to the means from three independent repeats, and the error bars show standard errors.

crp deletion adversely affects growth at low temperature.

The effect of crp deletion on growth at low temperature was investigated by monitoring the growth of both M182 and M182Δcrp cells with and without pDU9 and pDCRP at 37 and 15°C. The saturation OD₆₀₀ (Fig. 6A and B) and CFU (Fig. 6C) of M182 cells were not affected much when grown at either 37 or 15°C, whereas M182Δcrp cells showed a significantly reduced saturation OD₆₀₀ (∼3-fold) and CFU (∼5-fold) at 15°C. Further, the introduction of multicopy crp (pDCRP), but not pDU9, complemented the cold-sensitive phenotype of M182Δcrp cells (as measured by the saturation OD₆₀₀). Taken together, these results indicated that the presence of CRP is important for cell growth at low temperature.

Fig. 6.

Growth curve plotted as [log₁₀(OD₆₀₀) + 2] versus time for M182 and M182Δcrp cells, with and without pDU9 (no CRP) and pDCRP (WT CRP) plasmids at 37°C (A) or 15°C (B). y axis values are scaled up by the addition of 2 to avoid negative log₁₀(OD₆₀₀) values. This will shift the scale of the y axis to positive values without changing the shape/nature of the curve. (C) Bar diagram showing the CFU for both M182 and M182 Δcrp cells at the saturation OD₆₀₀ at both 37 and 15°C.

DISCUSSION

Our earlier study established that cspE, a member of the cspA family of cold shock proteins in E. coli, is an early cold-inducible protein (42). The nucleic acid melting ability and transcription antiterminator activity of CspE have been reported to be critical for growth at low temperature (35). Here, we show that CRP, a global regulator involved in sugar metabolism, regulates cspE transcription in E. coli. The sequence analysis of the cspE upstream region showed a putative CRP binding site. Results from both the transcription and translation fusion assays showed that CRP regulates P_cspE transcription. In addition, this regulation was found to be proportional to the crp copy number (Fig. 1). Further, the demonstration of CRP binding to the target site, centered at −61.5, in P_cspE (Fig. 2) strongly suggested a direct interaction of CRP with P_cspE. The partial deletion of this target site, leading to the loss of CRP binding (Fig. 3), as well as the use of CRP mutants having defective DNA binding domains (Fig. 4A) failed to activate P_cspE transcription, indicating that the binding of CRP to the target site in P_cspE is a prerequisite for this regulation. Interestingly, the partial deletion/mutation in the CRP target site resulted in reduced P_cspE activity to a level (Fig. 3E) which is much lower than that present in the crp deletion mutant, suggesting the involvement of a secondary regulatory factor.

The CRP target site location (−61.5) in P_cspE is similar to that of P_lac (−60.5), one of the best-characterized class I type CRP-dependent promoter (39). The target site in P_lac has 9/10 positions that are similar to those of the consensus (39), while the target site at P_cspE matches the consensus at 7/10 positions in the core binding motifs (Fig. 1B). Nonetheless, the nucleotide positions at 5, 7, and 8 in the first half and the symmetrically related positions (15, 16, 18) in the second half, reported to be critical for CRP binding (28), are conserved in the target site in P_cspE. Although the presence of both core motifs was found to be necessary for CRP to bind to the target site in P_cspE (Fig. 3A), an intact second core motif was shown to be more important for CRP-P_cspE interaction (Fig. 3B and C). Additionally, it should be noted that the sequence of the second core motif is closer to the consensus (4/5 matching positions) than the first core motif (3/5 matching positions). Moreover, the nucleotide position 6 in the first half and 17 in the second half, which are critical for the generation of two primary kinks on each side of the dyad axis of the CRP-DNA complex (32, 40), are conserved only in the second half of the site in P_cspE.

Different activating regions of CRP (AR1, AR2, and AR3) have been identified from a collection of positive-control CRP mutants that showed interaction with DNA but failed to stimulate RNAP and, as a result, failed to activate specific CRP-dependent promoters (class I and II) (7, 21). A defect in the AR1 domain of CRP fails to activate both class I and II promoters, while an improved AR3 domain reverses the effect of a defective AR1 domain and enhances the activity for class II promoters only (5). In vivo, P_cspE activity required only the AR1 domain of CRP (Fig. 4B), and the improved AR3 domain did not compensate for the defective AR1 domain for P_cspE activation (Fig. 4C). In addition, the improved AR3 domain did not enhance P_cspE activation; rather, it caused an ∼70% decrease. Similarly, an improved AR3 domain (K52N) caused an ∼30 to 50% decrease in the activation of the CC (+20) pmelR promoter, an artificial class I promoter (5, 44, 45). All of these observations confirmed that CRP interacts with P_cspE by the typical class I mechanism.

Certain determinants of the α-CTD of RNAP are required for transcription at both class I and II CRP-dependent promoters, such as lacP1, CC (−61.5) (38), and acsP2 (4), while the α-NTD of RNAP contributes only to the transcription from class II promoters, such as CC (−41.5) (31). In line with our earlier observation that P_cspE is a class I promoter, the transcription from P_cspE required various determinants in α-CTD but not in α-NTD (Fig. 5). The 287 determinant, which is critical for α-CTD interaction with the AR1 domain of CRP (6, 38), was found to be important for P_cspE transcription. Further, the 265 determinant of α-CTD, which is important for α-CTD-UP element interaction, also was shown to contribute to P_cspE transcription. The A/T-rich sequence just upstream of the −35 element (GTAAAAGTTA) in P_cspE could act as a potential UP element, contributing to transcription in a CRP-independent manner, as seen for P_rrnB (37) and P_lac (10). In line with this, the R265A substitution in α-CTD caused a 50% decrease in P_cspE activity in the presence of the plasmid control (pLG339 with no crp) (data not shown), indicating that the R265 residue indeed contributes to CRP-independent transcription from P_cspE. Interestingly, unlike a typical class I promoter (38), the substitution at residue 261 did not affect P_cspE transcription.

Some of the major perturbations encountered by the cold-shocked cells are the stabilization of nucleic acid secondary structures, decreased membrane fluidity (affecting nutrient transport), and reduced metabolism (12, 22, 33). CspE (along with other CSPs) is proposed to facilitate translation and transcription by resolving nucleic acid secondary structures at low temperatures, thereby aiding in the cold adaptation process (3, 25, 35, 46). This study shows, for the first time, that CRP regulates cspE, a cold shock gene in E. coli, by directly interacting with P_cspE and RNA polymerase by a typical class I mechanism. Further, our results that the deletion of CRP exhibited a growth defect at low temperatures (Fig. 6) suggest that CRP plays an important role in cold adaptation and growth at low temperature. Interestingly, a large number of genes involved in the nutrient transport and metabolic pathways are regulated by CRP (18) and also have been shown to be differentially regulated during a downshift in temperature (16, 34). Overall, this study paves a new path for exploring and understanding the cross-talk between cellular metabolic regulation and growth at low temperature.